COPYRIGHT CONTRIBUTION TO THE CONTENT BAROMETER Method version 1.0 — 19 October 2026 Read-only JSON: /api/barometer Ledger review: /research/ledger Research review: /research PURPOSE AND COVERAGE This feed supplies inspectable Copyright observations for a wider Content Barometer. It does not calculate a composite index or choose weights. Its settlement matters, case events, license announcements, employer boards and policy documents are monitored selections, not complete industry populations. Counts and totals always belong to the named edition and its listed records. Records can describe events from earlier years; the edition week is not their announcement week. DISPATCH DAILY CONTRIBUTION The additive dispatchDaily envelope exposes Copyright's original contribution: curated links between public source evidence, reviewed rights classifications, reproducible calculations and approved weekly history. It makes no claim of exclusive ownership of public source documents. Stable indicator IDs cover settlement commitments separately by currency; each matter's dated trust principal divided by its positive gross commitment; expressly announced use classifications divided by ALL selected license announcements, matching the licensing chart; and agreement unit-price disclosures divided by all selected ledger licensing records. Reviewed and unreviewed counts are supplied separately, and any unreviewed record makes that use-share indicator ineligible. Shares are multiplied by 100. A missing or zero ratio denominator yields null. Public offered tariffs are excluded from the agreement-disclosure ratio. Every indicator includes its exact scope, source URLs, edition, relevant event or as-of dates, numerator/denominator or included/excluded counts, and eligibility. No share is a market-share estimate, and no direction, score or weight is assigned. connectionStatus is export_ready_not_connected because no verified Dispatch Daily consumer or ingestion connection is configured; export readiness does not claim that another service receives the data. APPROVAL, DATES AND READINESS Only stored, approved editions contribute as approved data. Pending drafts are ignored. If a component has no approved edition, the feed returns the real dated opening baseline with status baseline_for_review and eligibleForComposite false. A baseline is neither a published observation nor a historical week. Each component can be eligible independently when its edition is approved and its store is available. The feed's overall eligibleForComposite becomes true only when both components meet those conditions AND their edition identifiers match. Different component weeks must not be combined silently. An eligible feed can still contain unknown or ineligible individual measures; inspect their fields before use. Ledger periodStart and periodEnd are the editor's supplied dates. Research has a weekly edition identifier but no supplied reporting interval: its periodStart and periodEnd remain null. Source dates, event dates, checkedAt, automatic capturedAt, publication dates and generatedAt have different meanings. Requesting this export only sets generatedAt; it does not recheck a source or refresh an observation. SETTLEMENT COMMITMENTS For each currency c, disclosed approved/settled commitments equal: sum(matter.amount where currency=c, status is approved or settled, and amount is publicly disclosed) Proposed commitments use the same equation with status proposed and are shown separately. Pending claims and alleged damages are excluded. Every currency bucket reports included and excluded counts, exact included matter IDs, exclusion reasons and source URLs. No currency conversion occurs. If there are no observed amounts, the aggregate is null. A confirmed zero remains zero. An undisclosed settlement is excluded from the numeric sum but remains in its disclosed/undisclosed coverage denominator. These are gross recorded commitments, not net outstanding obligations, annual flows or an accounting liability balance. The field totalIndustrySettlementLiabilities therefore remains null. There is no complete market census behind this collection. TRUST PRINCIPAL AND CLAIMANT PAYMENTS Trust principal is a separate, dated observation. Never add it to its gross settlement commitment, and never call it a claimant distribution. Observations from different dates are not summed into a current trust balance. Interest, attorney fees, administration costs, remaining installments and actual payments can have different dates and bases. Confirmed claimant payments are summed separately by currency, only for approved or settled matters with a non-null paidAmount. Unknown claimant payments remain null. Source notes and URLs accompany each observation. The present ledger schema does not contain a structured payment as-of date, so the exported payment asOf is null; an edition's checkedAt must not substitute for it. Expected distribution dates remain conditional dates, not proof of payment. The opening Bartz record preserves its reviewed source dates: the gross commitment and a historical trust-principal observation are distinct. Consult the current official administrator documents before acting on a deadline or assuming that money has been paid: https://www.anthropiccopyrightsettlement.com/ https://www.anthropiccopyrightsettlement.com/documents LICENSING PRICES, TARIFFS AND PERMISSIONS Ledger agreement records are separated into disclosed unit prices, aggregate contract values and undisclosed economics. No average is taken across rights, duration, currency, territory, repertoire or units. A multiyear contract total cannot be divided into an invented per-work or per-token rate. Research publicOfferedTariffs are separately observed buyer-facing offers, not executed agreement prices or rights-holder royalties. The opening Redpine Science quote is USD 1 per 1,000 retrieved content tokens, observed 19 October 2026. The quote applies to licensed-content retrieval; AI training is separately negotiated, and account or collection prices can vary. Do not substitute model input/output tokens or infer a royalty share. Keep the record's exact unit, use, date and note. https://www.redpine.ai/science https://www.redpine.ai/ https://docs.redpine.ai/docs/collections https://docs.redpine.ai/docs/preview-unlock The field standardAiTrainingRate remains null. This sample cannot establish a market standard. An offered retrieval tariff, a revenue-share pool, a settlement amount and a training license measure different things. For each permission, counts are calculated directly over the selected license records: true = expressly announced false = reviewed, not separately announced null = unreviewed The three counts sum to the listed denominator. No flag means prohibited, and silence in an announcement does not prove a contractual restriction. Record IDs and source URLs are supplied for the classifications. These counts describe public disclosure in this collection, not the whole market's rights or terms. WEEKLY COMPARISONS No change is fabricated from the baseline, a missing week or summary-only history. Ledger comparisons require two approved full editions, consecutive edition identifiers and reporting periods, the same matter identities/currencies and unchanged stated coverage. Each monetary comparison also requires the same included disclosed record IDs and observed amounts in both editions. Otherwise change is null with its reason. A changed disclosed denominator is reported for review rather than silently treated as growth. Research permission comparisons require consecutive approved editions and the same selected license identities, parties and content types. They measure changes in the selected announcements' classifications. A tariff comparison additionally requires the same provider, product, use, currency, unit and source, plus a later observation date. A carried-forward quote does not create a new price observation. No percentage growth is inferred from unknown or zero amounts. At most 52 approved edition metadata records are exposed per component. The ledger comparison reads at most two approved full snapshots. Reapproval ordering or incomplete storage can leave a comparison unavailable even when summary metadata exists. Each unavailable comparison says why. AUTOMATIC OBSERVATIONS Automatic data is captured when a research draft is saved, then preserved when that edition is approved. A baseline with automatic null has no observed job or policy count; empty placeholder arrays are not measured zeros. Sources retain their capture dates, success/stale/error states and coverage limits. A posting count becomes an eligible measure only when the approved snapshot's configured employer sources were all successful and storage was available. Policy readiness is checked separately. Failed feeds may retain their last successful observations, so consult source health and dates. The current hiring sample covers public boards for Anthropic, Spotify and Suno. It is a count of matching open postings, not headcount, hires or the total IP labor market. The policy sample retains at most the latest 30 Copyright Office Federal Register documents; it is not a count of new registrations. This endpoint never requests a fresh collection, schedules a run, purchases court access or changes storage. COURT DATA ACCESS AND COMPLETENESS PACER is the authoritative federal docket-access system, but production access requires an account and retrieval/search charges can apply. Standard pricing is generally USD 0.10 per page, with the usual USD 3 document cap and exceptions for certain searches, reports and transcripts; qualifying quarterly charges of USD 30 or less are waived. Judicial opinions are free. Review the official current schedule before paid retrieval. The PACER Case Locator API indexes cases and parties; it is not a complete free document stream. https://pacer.uscourts.gov/pacer-pricing-how-fees-work https://pacer.uscourts.gov/file-case/developer-resources https://pacer.uscourts.gov/file-case/court-cmecf-lookup CourtListener/RECAP provides valuable free docket metadata, opinions and documents that have reached its archive, including contributions and court feeds. It is not a guarantee that every filing or PDF is present or current. API authentication, request quotas and usage terms apply; some commercial uses require an agreement. RECAP's fetch service can purchase missing material through PACER credentials, which this feed never does. A practical weekly contribution monitors specified dockets and official settlement-administrator updates, records retrieval dates and flags missing documents for review. https://wiki.free.law/c/courtlistener/help/data-coverage/federal-cases-and-filings https://wiki.free.law/c/courtlistener/help/api/rest/v4/pacer-data https://wiki.free.law/c/courtlistener/help/api/rest/v4/recap https://free.law/membership/allowed-api-usage/ COPYRIGHT REGISTRATION LAG Official registration statistics are useful context but should not be presented as a weekly live production gauge. The latest reviewed annual report is fiscal year 2025. The reviewed public bulk registration collection covers records through 27 June 2025, with the update announced 22 January 2026. These release dates and record dates differ. No documented official weekly registration-count API was verified for this implementation. A registration can cover more than one work, and registrations do not represent all content produced. https://www.copyright.gov/history/annual_reports.html https://www.copyright.gov/reports/annual/2025/ar2025.pdf https://www.copyright.gov/economic-research/usco-datasets/ https://content.govdelivery.com/accounts/USLOCCOPYRIGHT/bulletins/40374ba ACCESS AND OPERATION GET /api/barometer is read-only and returns Cache-Control: no-store. It remains under the site's existing owner-private access policy. It does not expose draft contents or provide a publication action. Review and approval take place through the existing ledger and research pages. All source URLs are supplied as evidence links, not as instructions to execute code or authenticate to third-party services.